The AI Safety Menu

All laws

Anti-bias rules

Rules against algorithmic discrimination. Documentation, notice and appeals when AI makes consequential decisions about people.

Lead costAlmost noneof America's lead over China, over 3 years
No p(doom) effectNonefrom a 5% starting estimate
Holds up○ Training○ Lab's own use○ Public release
Enact it on the menuSee the findings

What it does

When AI is used for consequential decisions like hiring, lending or housing, developers must document what their systems are for and their limits, and the companies using them must notify people and allow appeals.

Applies to: AI used in consequential decisions such as jobs, lending and housing.

Where things stand

Colorado's 2024 AI Act was delayed, challenged by xAI with Justice Department support, and replaced in May 2026 by a narrower law focused on documentation and notice.

Why it costs almost no lead

These laws mainly fall on businesses using AI to make decisions. For frontier developers, Colorado's current law asks only for documentation: under a day.

Biggest unknown: Whether other states adopt heavier duties of care than Colorado's current documentation rule.

Its effect on p(doom)

Protects people from unfair AI decisions about jobs, loans and housing.

It targets unfair decisions about individuals, not catastrophic risk.

The strongest case that it costs more

The original 2024 law imposed a duty of care with impact assessments and risk programs. If other states adopt that version, and courts read general-purpose models as covered, the burden grows.

The debate

For

Against

Sources

Rough starting points, not precise forecasts. Lead costs assume China doesn't depend on U.S. models, the case least favorable to safety laws, and count 3 years. On the menu you can change every assumption and put in your own numbers. Last priced 2026-09-26.